Canada

This paper studies the impact of layer normalization (LayerNorm) on zero-shot translation (ZST). Recent efforts for ZST often utilize the Transformer architecture as the backbone, with LayerNorm at the input of layers (PreNorm) set as the default. However, Xu et al. (2019) has revealed that PreNorm carries the risk of overfitting the training data. Based on this, we hypothesize that PreNorm may overfit supervised directions and thus have low generalizability for ZST. Through experiments on OPUS, IWSLT, and Europarl datasets for 54 ZST directions, we demonstrate that the original Transformer setting of LayerNorm after residual connections (PostNorm) consistently outperforms PreNorm by up to 12.3 BLEU points. We then study the performance disparities by analyzing the differences in off-target rates and structural variations between PreNorm and PostNorm. This study highlights the need for careful consideration of the LayerNorm setting for ZST.

&lt;iframe src=&quot;https://app.sli.do/event/iHwnn5G9xYwcAmSbeKA4Au/embed/polls/6612fc62-9005-4727-8b04-ad01a5e3c7e6&quot; width=&quot;300&quot; height=&quot;400&quot;&gt;&lt;/iframe&gt;

ACL 2023

Exploring the Impact of Layer Normalization for Zero-shot Neural Machine Translation

layer-wise language recognition

zero-shot neural machine translation

post layer normalization

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
<br>The conference will be held in Toronto, Canada, July 9-14, 2023.<br><br>
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience.<br>
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**
<br><br><br><br>

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

This paper studies the impact of layer normalization (LayerNorm) on zero-shot translation (ZST). Recent efforts for ZST often utilize the Transformer architecture as the backbone, with LayerNorm at the input of layers (PreNorm) set as the default. However, Xu et al. (2019) has revealed that PreNorm carries the risk of overfitting the training data. Based on this, we hypothesize that PreNorm may overfit supervised directions and thus have low generalizability for ZST. Through experiments on OPUS, IWSLT, and Europarl datasets for 54 ZST directions, we demonstrate that the original Transformer setting of LayerNorm after residual connections (PostNorm) consistently outperforms PreNorm by up to 12.3 BLEU points. We then study the performance disparities by analyzing the differences in off-target rates and structural variations between PreNorm and PostNorm. This study highlights the need for careful consideration of the LayerNorm setting for ZST.

<iframe src="https://app.sli.do/event/iHwnn5G9xYwcAmSbeKA4Au/embed/polls/6612fc62-9005-4727-8b04-ad01a5e3c7e6" width="300" height="400"></iframe>

poster

Multilingual Knowledge Graph Completion (KGC) aims to predict missing links with multilingual knowledge graphs. However, existing approaches suffer from two main drawbacks: (a) alignment dependency: the multilingual KGC is always realized with joint entity or relation alignment, which introduces additional alignment models and increases the complexity of the whole framework; (b) training inefficiency: the trained model will only be used for the completion of one target KG, although the data from all KGs are used simultaneously. To address these drawbacks, we propose a novel multilingual KGC framework with language-sensitive multi-graph attention such that the missing links on all given KGs can be inferred by a universal knowledge completion model. Specifically, we first build a relational graph neural network by sharing the embeddings of aligned nodes to transfer language-independent knowledge. Meanwhile, a language-sensitive multi-graph attention (LSMGA) is proposed to deal with the information inconsistency among different KGs. Experimental results show that our model achieves significant improvements on the DBP-5L and E-PKG datasets.

<iframe src="https://app.sli.do/event/iHwnn5G9xYwcAmSbeKA4Au/embed/polls/767e1479-9c6f-476e-b4f3-284f27f4938f" width="300" height="400"></iframe>

Multilingual Knowledge Graph Completion with Language-Sensitive Multi-Graph Attention

In recent years, pre-trained Multilingual Language Models (MLLMs) have shown a strong ability to transfer knowledge across different languages. However, given that the aspiration for such an ability has not been explicitly incorporated in the design of the majority of MLLMs, it is challenging to obtain a unique and straightforward explanation for its emergence. In this review paper, we survey literature that investigates different factors contributing to the capacity of MLLMs to perform zero-shot cross-lingual transfer and subsequently outline and discuss these factors in detail. To enhance the structure of this review and to facilitate consolidation with future studies, we identify five categories of such factors. In addition to providing a summary of empirical evidence from past studies, we identify consensuses among studies with consistent findings and resolve conflicts among contradictory ones. Our work contextualizes and unifies existing research streams which aim at explaining the cross-lingual potential of MLLMs. This review provides, first, an aligned reference point for future research and, second, guidance for a better-informed and more efficient way of leveraging the cross-lingual capacity of MLLMs.

<iframe src="https://app.sli.do/event/iHwnn5G9xYwcAmSbeKA4Au/embed/polls/51752866-0361-4eff-9d08-e0f3fa31c522" width="300" height="400"></iframe>

Towards a Common Understanding of Contributing Factors for Cross-Lingual Transfer in Multilingual Language Models: A Review

Current neural semantic parsers take a supervised approach requiring a considerable amount of training data which is expensive and difficult to obtain. Thus, minimizing the supervision effort is one of the key challenges in semantic parsing. In this paper, we propose the Retrieval as Ambiguous Supervision framework, in which we construct a retrieval system based on pretrained language models to collect high-coverage candidates. Assuming candidates always contain the correct ones, we convert zero-shot task into ambiguously supervised task. To improve the precision and coverage of such ambiguous supervision, we propose a confidence-driven self-training algorithm, in which a semantic parser is learned and exploited to disambiguate the candidates iteratively. Experimental results show that our approach significantly outperforms the state-of-the-art zero-shot semantic parsing methods.

<iframe src="https://app.sli.do/event/iHwnn5G9xYwcAmSbeKA4Au/embed/polls/2614633d-497e-4887-bd5e-c91f78755e91" width="300" height="400"></iframe>

Ambiguous Learning from Retrieval: Towards Zero-shot Semantic Parsing

Relation Extraction (RE) is a task that identifies relationships between entities in a text, enabling the acquisition of relational facts and bridging the gap between natural language and structured knowledge. However, current RE models often rely on small datasets with low coverage of relation types, particularly when working with languages other than English.In this paper, we address the above issue and provide two new resources that enable the training and evaluation of multilingual RE systems.First, we present SRED<sup>FM</sup>, an automatically annotated dataset covering 18 languages, 400 relation types, 13 entity types, totaling more than 40 million triplet instances. Second, we propose RED<sup>FM</sup>, a smaller, human-revised dataset for seven languages that allows for the evaluation of multilingual RE systems. To demonstrate the utility of these novel datasets, we experiment with the first end-to-end multilingual RE model, mREBEL, that extracts triplets, including entity types, in multiple languages. We release our resources and model checkpoints at [https://www.github.com/babelscape/rebel](https://www.github.com/babelscape/rebel).

<iframe src="https://app.sli.do/event/jQes3hFZo3TYjGXEHm6w95/embed/polls/f5e597e7-2077-43e2-b3a7-291d1447e966" width="300" height="400"></iframe>

RED<sup>FM</sup>: a Filtered and Multilingual Relation Extraction Dataset

Pre-trained transformer-based models have recently shown great performance when applied to Named Entity Recognition (NER). As the complexity of their self-attention mechanism prevents them from processing long documents at once, these models are usually applied in a sequential fashion. Such an approach unfortunately only incorporates local context and prevents leveraging global document context in long documents such as novels, which might hinder performance. In this article, we explore the impact of global document context, and its relationships with local context. We find that correctly retrieving global document context has a greater impact on performance than only leveraging local context, prompting for further research on how to better retrieve that context.

<iframe src="https://app.sli.do/event/jQes3hFZo3TYjGXEHm6w95/embed/polls/2ab10a0d-d613-48da-ade5-4cdead11dbf1" width="300" height="400"></iframe>

The Role of Global and Local Context in Named Entity Recognition

Recent open-domain TableQA models are typically implemented as retriever-reader pipelines. The retriever component is usually a variant of the Dense Passage Retriever, which computes the similarities between questions and tables based on a single representation of each.These fixed vectors can be insufficient to capture fine-grained features of potentially very big tables with heterogeneous row/column information. We address this limitation by 1) applying late interaction models which enforce a finer-grained interaction between question and table embeddings at retrieval time. In addition, we 2) incorporate a joint training scheme of the retriever and reader with explicit table-level signals, and 3) embed a binary relevance token as a prefix to the answer generated by the reader, so we can determine at inference time whether the table used to answer the question is reliable and filter accordingly. The combined strategies set a new state-to-the-art performance on two public open-domain TableQA datasets.

<iframe src="https://app.sli.do/event/aL2QM8dUtT8fzBuK3envdP/embed/polls/d27c199e-84b1-4828-851b-eb07463d61ca" width="300" height="400"></iframe>

LI-RAGE: Late Interaction Retrieval Augmented Generation with Explicit Signals for Open-Domain Table Question Answering

Social science and psycholinguistic research have shown that power and status affect how people use language in a range of domains. Here, we investigate a similar question in a large, distributed, consensus-driven community with little traditional power hierarchy -- the Internet Engineering Task Force (IETF), a collaborative organisation that designs internet standards. Our analysis based on lexical categories (LIWC) and BERT, shows that participants' levels of influence can be predicted from their email text, and identify key linguistic differences (e.g., certain LIWC categories, such as "WE"} are positively correlated with high-influence). We also identify the differences in language use for the same person before and after becoming influential.

<iframe src="https://app.sli.do/event/aL2QM8dUtT8fzBuK3envdP/embed/polls/37410325-95b0-4c83-a05f-cbf569866665" width="300" height="400"></iframe>

Tracing Linguistic Markers of Influence in a Large Online Organisation

Link Prediction on Hyper-relational Knowledge Graphs (HKG) is a worthwhile endeavor. HKG consists of hyper-relational facts (H-Facts), composed of a main triple and several auxiliary attribute-value qualifiers, which can effectively represent factually comprehensive information. The internal structure of HKG can be represented as a hypergraph-based representation globally and a semantic sequence-based representation locally. However, existing research seldom simultaneously models the graphical and sequential structure of HKGs, limiting HKGs' representation. To overcome this limitation, we propose a novel Hierarchical Attention model for HKG Embedding (HAHE), including global-level and local-level attention. The global-level attention can model the graphical structure of HKG using hypergraph dual-attention layers, while the local-level attention can learn the sequential structure inside H-Facts via heterogeneous self-attention layers. Experiment results indicate that HAHE achieves state-of-the-art performance in link prediction tasks on HKG standard datasets. In addition, HAHE addresses the issue of HKG multi-position prediction for the first time, increasing the applicability of the HKG link prediction task. Our code is publicly available.

<iframe src="https://app.sli.do/event/aL2QM8dUtT8fzBuK3envdP/embed/polls/7a23dfae-7008-421b-a85d-8f280ab788fe" width="300" height="400"></iframe>

HAHE: Hierarchical Attention for Hyper-Relational Knowledge Graphs in Global and Local Level

Keeping track of how states of entities change as a text or dialog unfolds is a key prerequisite to discourse understanding. Yet, there have been few systematic investigations into the ability of large language models (LLMs) to track discourse entities. In this work, we present a task probing to what extent a language model can infer the final state of an entity given an English description of the initial state and a series of state-changing operations. We use this task to first investigate whether Flan-T5, GPT-3 and GPT-3.5 can track the state of entities, and find that only GPT-3.5 models, which have been pretrained on large amounts of code, exhibit this ability. We then investigate whether smaller models pretrained primarily on text can learn to track entities, through finetuning T5 on several training/evaluation splits. While performance degrades for more complex splits, we find that even when evaluated on a different set of entities from training or longer operation sequences, a finetuned model can perform non-trivial entity tracking. Taken together, these results suggest that language models can learn to track entities but pretraining on text corpora alone does not make this capacity surface.

<iframe src="https://app.sli.do/event/6qSnTPZvioRLfFJpZUjhCw/embed/polls/66956f35-385a-41bb-8a73-a0f7dab4e272" width="300" height="400"></iframe>

Entity Tracking in Language Models

Recent abstractive conversation summarization systems generally rely on large-scale datasets with annotated summaries. However, collecting and annotating these conversations can be a time-consuming and labor-intensive task. To address this issue, in this work, we present a sub-structure level compositional data augmentation method, \textsc{Compo}, for generating diverse and high-quality pairs of conversations and summaries. Specifically, \textsc{Compo} first extracts conversation structures like topic splits and action triples as basic units. Then we organize these semantically meaningful conversation snippets compositionally to create new training instances.Additionally, we explore noise-tolerant settings in both self-training and joint-training paradigms to make the most of these augmented samples. Our experiments on benchmark datasets, SAMSum and DialogSum, show that \textsc{Compo} substantially outperforms prior baseline methods by achieving a nearly 10\% increase of ROUGE scores with limited data. Code is available at \texttt{https://github.com/ozyyshr/Compo}.

<iframe src="https://app.sli.do/event/6qSnTPZvioRLfFJpZUjhCw/embed/polls/353c70e7-bdef-4056-83ae-08a031a8ba7f" width="300" height="400"></iframe>

Downloads

Next from ACL 2023

Multilingual Knowledge Graph Completion with Language-Sensitive Multi-Graph Attention

Similar lecture

On the Expressivity Role of LayerNorm in Transformers' Attention